Goldman Sachs AI Adoption: Rebuilding the Productivity Measurement Framework

Akhil joined Goldman Sachs's India AI adoption program as one of the lead program managers. The mandate came from the New York office and was handed to the India CEO; as one of the Chief of Staffs to the India CEO, Akhil was assigned end-to-end program ownership.

The early phase was tracking-focused: ensuring adoption numbers were being captured across divisions. Adoption reached 90 to 95%, and on the surface the program looked like a success. But when Akhil and the team started examining productivity outcomes, the numbers did not follow. It was straightforward to measure how frequently employees were using AI tools, but there was no reliable method to determine whether that usage was producing better work.

That gap required building a new measurement framework from scratch. Akhil worked across divisions to identify metrics that could connect AI usage to actual output quality and efficiency, not just frequency of use. That involved talking to senior and junior stakeholders across the firm, understanding how different teams defined productivity, and designing a tracking approach that could work across functions.

The second challenge was organizational. When the team began rolling out the new framework, several division leaders pushed back, interpreting the initiative as an attempt to monitor and control their junior staff. Akhil recalibrated the approach: rather than running the program top-down, he brought division leaders in as champions, giving them ownership of implementation within their teams and incorporating their input into the methodology. That shift moved the program from a compliance exercise to a firm-wide initiative with distributed ownership.